基於分散式情感向量之深層神經網路中文對話生成研究

Refined Distributed Emotion Vector Representation for Deep Neural Network to Generate Chinese Emotional Conversation

邢晏純(Yan-Chun Hsing)

指導教授 ： 張詠淳

臺北醫學大學/管理學院/大數據科技及管理研究所/碩士(2019年)

摘要

隨著社群網站記錄的文字越來越多，越能提供對話技術的成長，系統不僅會傳送罐頭訊息，更能夠理解語意、擷取情緒資訊，進而與人類進行有效的溝通。本文利用序列到序列法結合情緒特徵方法解決短文本對話(Short text conversation, STC)中，生成具有情感回覆之問題。其中也運用了非監督式學習方法，自動評估情緒向度(W2VA)，取代往往需要透過人工標註的情緒語料，也提高情緒分類的精確度，並且結合卷積神經網路與遞歸神經網路改善回覆文字，使每句回覆皆具有適當的內容和充足的情感，本實驗更建立一個GEST模型(Generated Emotional Short Text model)，其為一具有人性化且有情感的中文對話系統，期望使用者透過此系統能夠拉近人機之間的互動關係，並且驗證了人類偏好於引入情感因素的對話模型。

關鍵字

情緒分析 ； 情緒向度 ； 對話系統 ； 序列到序列

並列摘要

With the increasing number of texts recorded on social media, the more we can provide the growth of dialogue technology. So that the system not only transmits lazy texts, but also understands semantics, extracts emotional information, and communicates effectively with humans. This paper solves the problem of short text conversation generates emotional response by using sequence to sequence combined with emotional feature method. It uses the method of automatic evaluation of emotional dimension (W2VA) to replace the emotional corpus that often needs to be manually annotated, and improve the accuracy of emotional classification and combined with CNN and RNN to improve the reply text so that each reply has appropriate content and emotion. This experiment also establishes a Generated Emotional Short Text model, which is a humanized and emotional Chinese conversation system. It is expected that users can bring the human-computer interaction closer through this system.

並列關鍵字

Sentiment Analysis ； Emotional Dimensions ； Dialogue System ； Seq2Seq

參考文獻

Cambria, E. (2016). Affective computing and sentiment analysis. IEEE Intelligent Systems, 31(2), 102-107.

Zhang, Y., Li, J., Song, Y., & Zhang, C. (2018). Encoding conversation context for neural keyphrase extraction from microblog posts. Paper presented at the Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers).

Lopez, M. M., & Kalita, J. (2017). Deep Learning applied to NLP. arXiv preprint arXiv:1703.03091.

Alm, C. O., Roth, D., & Sproat, R. (2005). Emotions from text: machine learning for text-based emotion prediction. Paper presented at the Proceedings of the conference on human language technology and empirical methods in natural language processing.

Agarwal, A., Xie, B., Vovsha, I., Rambow, O., & Passonneau, R. (2011). Sentiment analysis of twitter data. Paper presented at the Proceedings of the Workshop on Language in Social Media (LSM 2011).